Morphometric analysis of relic landslidesusing detailed landslide distribution maps:Implications for forecasting travel distanceof future landslides

著者 Hattanji Tsuyoshi, Moriwaki Hiromujournal orpublication title

Geomorphology

volume 103number 3page range 447-454year 2009-02権利 (C) 2008 Elsevier B.V.URL http://hdl.handle.net/2241/101655

doi: 10.1016/j.geomorph.2008.07.009

1

Morphometric analysis of relic landslides using detailed landslide distribution 1

maps: Implications for forecasting travel distance of future landslides 2

3

Tsuyoshi Hattanji*a, Hiromu Moriwakib 4

5 a Graduate School of Life and Environmental Sciences, University of Tsukuba, 305–6

8572, Japan 7 b National Research Institute for Earth Science and Disaster Prevention (NIED), 8

305–0006, Japan 9

* Corresponding author. Tel: +81 29 853 5687; fax: +81 29 853 6879 10

E-mail address: hattan@geoenv.tsukuba.ac.jp (T. Hattanji) 11

12

Abstract 13

14

This study analyzed the morphometry of relic landslides in four mountainous areas in 15

Japan. Using detailed landslide maps issued by the National Research Institute for 16

Earth Science and Disaster Prevention of Japan, the mobility of 338 relic landslides 17

was evaluated based on the H/L ratio, i.e., the equivalent coefficient of friction. The 18

H/L ratio strongly correlates with the initial slope, tanθr (R2 = 0.78–0.88). The H/L 19

ratio and tanθr of recent disaster-causing landslides within the past few decades were 20

also measured in each investigated area. The data of recent landslides correspond to 21

the 95% lower prediction limit of the tanθr–H/L regression line for relic landslides. 22

This result implies that morphometric analysis of relic landslides around an unstable 23

mass allows for forecasting of travel distance of the unstable mass. 24

25

Keywords: Landslide map, Morphometric analysis, Mobility, Travel distance, 26

Equivalent coefficient of friction 27

28

2

1. Introduction 29

30

Landsliding is a major geomorphic process affecting landscape evolution in 31

mountainous regions (Roering et al., 2005), and causing catastrophic disasters. Since 32

natural and human impacts may reactivate some relic landslides (Ost et al., 2003; 33

Chigira and Yagi, 2006), a landslide map based on aerial photographic interpretation 34

is useful for understanding the possibility of landslide reactivation in the future. 35

Analysis of landslide susceptibility using a landslide map has been developed with a 36

multivariate statistical approach (Carrara et al., 1991, 2003; Ayalew and Yamagishi, 37

2005) and the analytic hierarchy process method (Yoshimatsu and Abe, 2006). 38

However, not only assessment of landslide susceptibility, but prediction of mobility is 39

necessary for mitigation of landslides. 40

Scheidegger (1973) and Hsü (1975) suggested an index of landslide mobility, H/L, 41

where H is the fall height and L is the horizontal length of an entire landslide. The 42

H/L ratio, which is equivalent to the gradient of the line from toe to top of a landslide, 43

is refereed to as the ‘equivalent coefficient of friction’ from the standpoint of 44

kinematics. Scheidegger (1973) and Hsü (1975) also indicated that the H/L ratio of a 45

rock avalanche decreases with increasing volume of the avalanching mass. 46

Corominas (1996) confirmed this size effect on the H/L ratio for other types of 47

landslides. Moriwaki (1987) revealed that the H/L ratio decreases with decreasing 48

initial slope for both natural and experimental landslides. Recent statistical analyses 49

for shallow landsliding also ensured the effects of topography on the H/L ratio (Finlay 50

et al., 1999; Hunter and Fell, 2003; Okura et al., 2003). 51

The above studies used various datasets of recent landslides, which may include 52

landslides under diverse local environmental conditions such as geologic, topographic 53

and climatic. Nevertheless, a landslide map provides H/L ratios and related 54

parameters for the target area. Few studies, however, quantitatively measured 55

3

landslide mobility based on a large data set from landslide maps. 56

Moriwaki and Hattanji (2002) obtained the mobility of relic landslides in three 57

landslide-prone areas by using detailed landslide maps issued by the National 58

Research Institute for Earth Science and Disaster Prevention of Japan (NIED). The 59

NIED landslide maps are based on the interpretation of 1:40 000 aerial photographs, 60

and distinctly show crowns, flanks, and displaced masses of moderate-scale relic 61

landslides (>100 m in width) on 1:50 000 topographic maps (Fig. 1). Moriwaki and 62

Hattanji (2002) reported that the H/L ratio strongly correlates with the initial slope of 63

the relic landslides, and they hypothesized that morphometric analysis of relic 64

landslides is useful for forecasting the travel distance of future landslides under 65

similar environmental conditions. 66

A problem remaining to be solved is the practicability of morphometric analysis to 67

forecasting. For the analysis of relic landslides, subsequent erosion may alter the 68

topography and affect the regression line of initial slope and H/L ratio accordingly. To 69

make the morphometric analysis more practical for forecasting, it is essential to 70

compare the outputs from the analysis of relic landslides with those of recent disaster-71

causing landslides. In addition, this previous study only focused on major landslide-72

prone areas where the landslide-area ratio is extremely high (20 – 30%). Recent 73

disaster-causing landslides, however, have occurred in areas with relatively low 74

landslide-area ratios (< 10%) as well as in major landslide-prone areas. It is necessary 75

to apply the morphometric analysis for more diverse geomorphic conditions. 76

(Fig. 1) 77

We analyze the morphometry of relic landslides in four mountainous areas in Japan 78

where landslide disasters have occurred within the past few decades (Fig. 2). Then we 79

compare results of the morphometric analysis of these relic landslides with those of 80

recent landslides in the same region. Finally, we discuss application of this analysis 81

using landslide maps to forecast the travel distance of future landslides. 82

4

(Fig. 2) 83

84

2. Investigated areas and recent landslide disasters 85

86

2.1. Description of the investigated areas 87

88

Table 1 shows the geologic, climatic, and geomorphic conditions of four investigated 89

areas: Hachimantai, Nagano, Gojo, and Ichinomiya. Each area covers 706.9 km2, 90

which is equivalent to a circle with a radius of 15 km. The climate of all the areas is 91

classified as humid temperate, though temperature and precipitation conditions vary 92

regionally. Snow accumulates over 2 m for the Hachimantai area, and less than 1 m 93

for the other areas. Planted and natural forests are the dominant vegetation in most of 94

these areas. 95

Hachimantai is an active volcanic area where geothermal activities around the 96

volcanoes stimulate landsliding. Structures of several old calderas also control the 97

distribution of large-scale landslides (Oyagi and Ikeda, 1998). These factors are the 98

cause of the high percentage of landslide area (23.1%). The Nagano area consists of 99

both uplifted mountains underlain by Neogene rocks and an inactive Quaternary 100

volcano. The percentage of landslide area increases up to 17.5% for the area 101

excluding a large floodplain of the Shinano River. The Gojo and Ichinomiya areas are 102

uplifted mountains underlain by various lithologies. Although both areas have 103

relatively low percentages of landslide area (3 – 6%), landslide disaster occurred in 104

each area during the past few decades. 105

(Table 1) 106

107

2.2. Descriptions of recent landslides 108

2.2.1. Sumikawa landslide – Hachimantai area 109

5

110

A landslide occurred at a headwater of the Akagawa River in the Hachimantai area on 111

May 11, 1997 (Fig. 3a). The displaced mass was composed of Quaternary volcanics, 112

including tuff and breccia. This site was previously identified as a relic landslide on 113

the landslide map ‘Hachimantai’ published in 1984 (Inokuchi, 1998). Matsuura et al. 114

(1998) revealed that groundwater recharge by snowmelt and heavy rainfall (110 mm) 115

on May 8 triggered the 1997 landslide. Steam explosions (the star symbol in Fig. 3a) 116

followed immediately after the landslide, and the displaced mass converted to a series 117

of debris avalanches that traveled 2 km downstream along the Akagawa River 118

(Chigira et al., 1998; Hayashi et al., 1998). 119

(Fig. 3) 120

121

2.2.2. Jizukiyama landslide – Nagano area 122

123

The Jizukiyama landslide occurred in a suburb of Nagano City in 1985. The 124

underlying geology is weathered Neogene rhyolite tuff. The Nagano Prefectural 125

Government office found small extension cracks on the slope in 1981, and installed 126

extensometers in May 1984. The creep rate was gradual in June 1985 but rapidly 127

accelerated after a thunderstorm that produced 58 mm of rainfall on July 20. The 128

unstable mass slid at 5:00 pm on July 26, and flowed into residential areas (Fig. 3b). 129

A southward flow involving a welfare facility caused 26 deaths, and two eastward 130

flows buried about 50 houses (Oyagi et al., 1986). 131

132

2.2.3. Wada landslide – Gojo area 133

134

The Wada landslide occurred at a valley-side slope of the Niu River in the Gojo area 135

on August 4, 1982 (Fig. 3c). The underlying geology is Mesozoic slate of the 136

6

Chichibu zone. The landslide was located at a section of a relic landslide (Okunishi 137

and Okuda, 1982). After heavy rainfall of 350 mm from August 1 to 3, a local 138

resident observed a small crack on the slope. The crack rapidly extended and two 139

landslides occurred at 2:00 and 8:15 am on August 4. The displaced mass temporally 140

dammed up the Niu River, causing flooding in a major residential area on the 141

opposite side of the river (Yonetani et al., 1983). 142

143

2.2.4. Fukuchi landslide – Ichinomiya area 144

145

The Fukuchi landslide occurred in the Ichinomiya area on September 13, 1976 (Fig. 146

3d). The landslide was located on a NW–SE fault zone, which borders sedimentary 147

rocks, volcanic rocks, and granodiorite. According to a tradition told in Ichinomiya 148

Town, a large-scale landslide occurred at the same place about 300 years ago 149

(Okunishi, 1982). From September 8 to 12, a typhoon brought 717 mm of rainfall. 150

The large-scale landslide occurred at 9:20 am on September 13, following a small 151

landslide at 6:30 am. The displaced mass transformed into an earth flow, which 152

buried a primary school and many houses. A local resident recorded the landslide 153

process from 9:05 to 9:30 in a sequence of photographs. Although most inhabitants 154

evacuated immediately before the large-scale landslide, three people were reported 155

missing due to the small landslide at 6:30 am (Oyagi et al., 1977; Okunishi, 1982). 156

157

3. Morphometric analysis of relic and recent landslides 158

159

We used 1:50 000 landslide maps issued by NIED (Shimizu and Oyagi, 1985; 160

Shimizu et al., 1984, 2003, 2004, 2005a,b) for morphometric analysis. To precisely 161

analyze landslide mobility, we selected relic landslides from NIED landslide maps 162

under the following conditions: (a) distinct main scarps, flanks, and outlines of 163

7

ruptured surfaces, (b) linear movement, and (c) area of ruptured surface greater than 164

0.01 km2. 165

To compare the mobility of the relic and recent landslides, it is necessary to select 166

relic landslides with environmental conditions similar to those of the recent landslides. 167

A simple method is to select relic landslides using the distance from recent landslides 168

D. Fig. 4 shows the relationship between D and the number of available relic 169

landslides or relative height in the investigated area. The topography of the 170

investigated areas becomes more diverse with increasing D as indicated by the 171

increase in relative height. In the case of the Gojo area, the investigated area for D = 172

20 km includes high-relief mountains (~ 1700 m a.s.l.), while the recent landslide 173

occurred in a low-relief hillslope (~ 400 m a.s.l.). In the Hachimantai and Ichinomiya 174

areas, the numbers of available relic landslide are very small for D < 10 km. 175

Consequently, we used relic landslides with D ≤ 15 km. The number of the landslides 176

is 73 for Hachimantai, 92 for Nagano, 93 for Gojo, and 80 for Ichinomiya. 177

(Fig. 4) 178

We followed the methodology of topographic measurement used in our previous 179

research (Moriwaki and Hattanji, 2002). The items measured were the horizontal 180

length and fall height of the entire landslide (L and H) as well as the horizontal length, 181

height, and area of ruptured surface (Lr, Hr, and Ar) (Fig. 5a). The terminology of the 182

measured items is based on the IAEG Commission on Landslides (1990). Although 183

landslide length is often defined by the length along a slope (Cruden and Varnes, 184

1996), we used horizontal length for convenience. Measuring Lr and Hr is difficult 185

because the outline of the ruptured surface around the landslide foot is generally 186

covered with displaced mass (Fig. 5a). We estimated the outline of the foot from the 187

extrapolation of the outline of the main scarp and flanks, and considered elevation of 188

the lowest ground surface along the extrapolated outline as the elevation of the foot 189

(Fig. 5b). 190

8

(Fig. 5) 191

We measured the morphometry of recent landslides with the same method (Table 192

2). The L and H measurements for the Sumikawa landslide include those of the debris 193

avalanche (the grey line in Fig. 3a), and L and H of the other landslides were 194

measured along their cross sections, as shown in Figs. 3b-d (i.e., B-B’, C-C’ and D-195

D’). We inferred the toe of the ruptured surface from the initial topography and the 196

slip surface in the cross sections. Yonetani et al. (1983) reported that the displaced 197

mass of the Wada landslide reached the opposite bank of the Niu River (toe 2 in Fig. 198

3c). The cross section steepened at toe 1 (Okunishi and Okuda, 1982, Fig. 3c), which 199

implies erosion of the displaced mass by fluvial processes after the landsliding. 200

(Table 2) 201

202

4. Results and Discussion 203

204

4.1 Effects of landslide size on H/L 205

206

Travel distance is essential for forecasting the area affected by landslides. As noted 207

before, Scheidegger (1973) and Hsü (1975) proposed that the H/L ratio decreases 208

with increasing volume of rock avalanche. However, volume data is not available on 209

the landslide map we used. Many studies show a clear power-law relationship 210

between volume and area of landslides (Innes, 1983; Guthrie and Evans, 2004), 211

implying a positive correlation between depth and volume. We assumed this 212

correlation and used the area of ruptured surface, Ar, as an indicator of size. Fig. 6 213

shows the H/L ratio plotted against logAr. A simple least-squared regression yielded 214

low coefficients of determination for all areas (0.08–0.26); however, they are 215

statistically significant (p < 0.01). 216

(Fig. 6) 217

9

Most of the relic landslides we analyzed had an Ar range of 0.01 to 1 km2, which is 218

narrower than that of some previous studies (Scheidegger, 1973; Hsü, 1975; 219

Corominas, 1996). In addition, the effect of the landslide area on H/L is less 220

straightforward than that of volume inferred in the previous studies. Even if the 221

volume of landslides was used, Okura et al. (2003) reported no correlation between 222

the volume and H/L for shallow landslides under a limited scale (103–104 m3). In 223

general, the size effect becomes less obvious under a limited range of size. Hsü 224

(1975) indicated that the H/L ratio decreases when the volume exceeds 5×106 m3. 225

This threshold is equivalent to Ar = 0.5 km2 for an average depth of 10 m, and would 226

be a reason for the stronger correlation in Hachimantai with larger landslides (Fig. 6). 227

Only 2% of all the relic landslides exceeds the threshold of Ar = 0.5 km2. Therefore, 228

we ignored the size effect for forecasting travel distance in the investigated areas. 229

However, the size effect may be a key factor if a target unstable mass has a larger size 230

than those of this study. 231

232

4.2 Effects of initial slope on H/L 233

234

Moriwaki (1987) and Okura et al. (2003) revealed a correlation between the 235

equivalent coefficient of friction and the initial slope of a landslide, i.e. slope of the 236

ruptured surface, tanθr (= Hr/Lr, Fig. 5a). Considering the kinematics of a landslide, 237

this correlation means that a gentler slope has a lower residual strength, which 238

determines the gradient of the displaced mass after landsliding. Fig. 7 indicates 239

definite positive correlations of H/L with tanθr for the relic landslides in the four areas. 240

This result is probably associated with long-term landscape evolution, although this is 241

a topic for future studies. Linear least-squares regression analysis gave the following 242

equation for each area: 243

244

10

Hachimantai: H/L = 0.77 tanθr + 0.040 (1a) 245

Nagano: H/L = 0.82 tanθr + 0.034 (1b) 246

Gojo: H/L = 0.91 tanθr + 0.035 (1c) 247

Ichinomiya: H/L = 0.83 tanθr + 0.055 (1d) 248

249

(Fig. 7) 250

These regression lines have much higher coefficients of determination (0.78–0.88) 251

than those of H/L on Ar (0.08–0.26). All the intercepts of Eq. (1) are significantly 252

different from null (p < 0.05). The slope of the regression line, however, indicates 253

average reduction of slopes before and after landsliding because the intercepts are 254

relatively small (0.03–0.05). In other words, a smaller slope of the regression line (Eq. 255

1) means higher mobility of landslides. Landslide mobility of Hachimantai is 256

definitely higher than that of Gojo. A Student’s t-test for the difference in the slopes 257

of these regression lines showed that the difference is statistically significant (p < 258

0.05). 259

The recent landslides were plotted below the regression lines of Eq. (1) for all the 260

four areas (squares in Fig. 7), meaning that the recent landslides had higher mobility 261

compared to the average of the relic landslides. This may reflect the type of landslides. 262

For example, the recent landslides in Hachimantai and Ichinomiya are flow-type 263

landslides (Fig. 3a,d), although the relic landslides in these areas include some other 264

types with lower mobility. 265

Another possible explanation of the difference between the recent and relic 266

landslides is erosion of the displaced mass after landsliding. For example, the toe of 267

the Wada landslide was eroded by flow of the Niu River after landsliding (Fig. 3c, 268

Yonetani et al., 1983). Although the downslope motion of a landslide leads to a 269

smaller H/L ratio than initial slope, tanθr, several relic landslides in all the areas have 270

H/L ratios greater than tanθr. Although errors of measurement or mapping might have 271

11

caused this problem, erosion after landsliding also increases the H/L ratio. 272

Topographic change by erosion after landsliding is probable in two cases: (1) the 273

displaced mass moves into a river or a high-order stream, and (2) age of landslide 274

initiation is old enough, to allow valley incision or headward erosion that affects 275

topography of the displaced mass. Therefore, lower prediction limits of the regression 276

lines for the relic landslides are more practical for understanding the behavior of 277

future landslides. 278

Accordingly, we calculated the 95% lower prediction limits for the regression lines 279

of Eq. (1). Since the statistical uncertainty in the slopes of Eq. (1) was negligible for a 280

tanθr of 0.1 to 0.8, the 95% limits were approximated to the following linear 281

equations (the thick solid lines in Fig. 7): 282

283

Hachimantai: H/L = 0.77 tanθr – 0.047 (2a) 284

Nagano: H/L = 0.82 tanθr – 0.040 (2b) 285

Gojo: H/L = 0.91 tanθr – 0.049 (2c) 286

Ichinomiya: H/L = 0.83 tanθr – 0.037 (2d) 287

288

The H/L ratios of the recent landslides agree well with Eq. (2) for the Hachimantai 289

and Ichinomiya areas, and are only slightly larger than those from Eq. (2) for the 290

other two areas (Fig. 7). Although the number of examples is limited, this result 291

indicates that morphometric analysis of relic landslides is useful for forecasting the 292

travel distance of future landslides under similar environmental conditions. 293

Regression analysis between tanθr and the H/L ratio for relic landslides enables us 294

to forecast a possible H/L ratio of an unstable mass. Requirements for this prediction 295

are the average slope of an unstable mass expected to slide and a detailed landslide 296

map, which separately outlines crowns, flanks, and displaced masses. Table 3 shows 297

the 95% lower prediction limits of the regression lines for D = 5, 10, and 15 km. For 298

12

the Nagano area where landslide density is higher than that of the other areas, the 299

95% lower prediction limit even for D = 5 km agrees well with the H/L of the recent 300

landslides. For the Hachimantai and Ichinomiya areas, however, the regression 301

analysis for D = 5 km is less significant due to the limited number of relic landslides. 302

This result indicates that at least 10 relic landslides are required for a confident 303

regression analysis. The appropriate D value for the analysis probably depends on 304

the density of available relic landslides and the variety in environmental conditions 305

(Fig. 4), although further studies are requied to discuss this issue. 306

(Table 3) 307

308

5. Conclusions 309

310

The present study explored the possibility of landslide maps for forecasting landslide 311

travel distance. We analyzed the mobility of relic landslides outlined in the NIED 312

landslide map in four landslide-prone areas, and examined the effects of size and 313

initial slope on the H/L ratio of landslides. Inverse correlation between the H/L ratio 314

and area of ruptured surface, Ar, was very weak, indicating a limited size effect under 315

the scale of landslides shown in the NIED landslide map (0.01 < Ar < 1 km2). Instead, 316

the H/L ratio strongly correlates with the initial slope, i.e. slope of the ruptured 317

surface, tanθr. On the tanθr–H/L charts, recent disaster-causing landslides were 318

plotted near the 95% lower prediction limit of the regression lines for the relic 319

landslides. Therefore, morphometric analysis of relic landslides in a given area 320

enables us to predict the possible travel distance of an unstable mass in the area. In 321

other words, detailed landslide maps may play an important role in forecasting 322

landslide travel distance. 323

324

Acknowledgements 325

13

We thank Y. Matsukura for his comments on an early draft of the manuscript, and Nel 326

Caine and one anonymous reviewer for their comments that improved the manuscript. 327

328

References 329

330

Ayalew, L., Yamagisghi, H., 2005. The application of GIS-based logistic regression 331

for landslide susceptibility mapping in the Kakuda-Yahiko Mountains, Central 332

Japan. Geomorphology 65, 15–31. 333

Carrara, A., Cardinali, M., Detti, R., Guzzetti, F., Pasqui, V., Reichenbach, P., 1991. 334

GIS techniques and statistical models in evaluating landslide hazard. Earth 335

Surface Processes and Landforms 16, 427–445. 336

Carrara, A., Crosta, G., Fratini, P., 2003. Geomorphological and historical data in 337

assessing landslide hazard. Earth Surface Processes and Landforms 28, 1125–338

1142. 339

Chigira, M., Yagi, H., 2006. Geological and geomorphological characteristics of 340

landslides triggered by the 2004 Mid Niigata prefecture earthquake in Japan. 341

Engineering Geology 82, 202– 221. 342

Chigira, M., Chiba, T., Matsuura, S., 1998. Landslide-triggered steam explosion and 343

debris flow at the Sumikawa Spa, Akita, Northern Japan, May 1997 (1) general 344

view and geological background. Landslide News 11, 6–8. 345

Corominas, J., 1996. The angle of reach as a mobility index for small and large 346

landslides. Canadian Geotechnical Journal 33, 260–271. 347

Cruden, D.M., Varnes, D.J., 1996. Landslide types and processes, In: Turner, A.K., 348

Schuster, R.L. (Eds.), Landslides: Investigation and Mitigation (Transportation 349

Research Council Special Report No. 247). National Academy Press, Washington 350

D.C., pp. 36–75. 351

Finlay, P.J., Mostyn, G.R., Fell, R., 1999. Landslide risk assessment: prediction of 352

14

travel distance. Canadian Geotechnical Journal 36, 556–562. 353

Guthrie, R.H., Evans, S.G., 2004. Analysis of landslide frequencies and characteristics 354

in a natural system, coastal British Columbia. Earth Surface Processes and 355

Landforms 29, 1321–1339. 356

Hayashi, S., Taniguchi, H., Chiba., T., 1998. Landslide-triggered steam explosion and 357

debris flow at the Sumikawa Spa, Akita, Northern Japan, May 1997 (2) 358

chronological records of events and steam explosions. Landslide News 11, 8–11. 359

Hoshino, M., Asai, K., 1997. The landslide disaster at Hachimantai-Sumikawa in 360

May 1997 (preliminary report). Journal of the Geographical Survey Institute 361

(Japan) 88, 28–40 (in Japanese). 362

Hsü, K.J., 1975. Catastrophic debris streams (Sturzstroms) generated by rockfalls. 363

Geological Society of America Bulletin 86, 129–140. 364

Hunter, G., Fell, R. 2003. Travel distance angle for "rapid" landslides in constructed 365

and natural soil slopes. Canadian Geotechnical Journal 40, 1123–1141. 366

IAEG Commission on Landslides, 1990. Suggested nomenclature for landslides. 367

Bulletin of International Association of Engineering Geology 41, 13–16. 368

Innes, J.L., 1983. Lichenometric dating of debris-flow deposits in the Scottish 369

Highlands. Earth Surface Processes and Landforms 8, 579–588. 370

Inokuchi, T., 1998. Landslide topography of the Sumikawa landslide. Journal of the 371

Japan Landslide Society 35(2), 11–19 (in Japanese with English Abstr.). 372

Matsuura, S., Asano, S., Okamoto, T., Takeuchi, Y., Nakamura, S., 1998. Estimation 373

of intensity of meltwater flowing out from the bottom of a snow pack prior to 374

Hachimantai-Sumikawa landslide and its probability analysis using AMeDAS 375

data set. Journal of the Japan Landslide Society 35(2), 20–28 (in Japanese with 376

English Abstr.). 377

Moriwaki, H., 1987. A prediction of the travel distance of a debris. Journal of the 378

Japan Landslide Society 24 (2), 10– 16 (In Japanese with English Abstr.). 379

15

Moriwaki, H., Hattanji, T., 2002. Morphometric analysis of landslides based on 380

landslide maps. Journal of the Japan Landslide Society 39, 244–252 (in Japanese 381

with English Abstr.). 382

Okunishi, K., 1982. Kinematics of large-scale landslides – A case study in Fukuchi, 383

Hyogo Prefecture, Japan. Transactions, Japanese Geomorphological Union 3, 41–384

56. 385

Okunishi, K., Okuda, S., 1982. Refuge from large scale landslides – A case study: 386

Nishiyoshino Landslide, Nara Prefecture, Japan. Journal Natural Disaster Science 387

4(2), 79–85. 388

Okura, Y., Kitahara, H., Kawanami, A., Kurokawa, U., 2003. Topography and volume 389

effects on travel distance of surface failure. Engineering Geology 67, 243–254. 390

Ost, L., Van Den Eeckhaut, M., Poesen, J., Vanmaercke-Gottigny, M.C., 2003. 391

Characteristics and spatial distribution of large landslides in the Flemish 392

Ardennes (Belgium), Zeitschrift für Geomorphologie N.F. 47, 329–350. 393

Oyagi, N., Ikeda, H., 1998. Landslide structure and regional perspective on 394

Sumikawa landslide at Hachimantai volcanic area, Northern Honshu, Japan. 395

Journal of the Japan Landslide Society 35(2), 1–10 (in Japanese with English 396

Abstr.). 397

Oyagi, N., Tanaka, K., Fukuzono, T., 1986. Report on the Jizukiyama landslide 398

disaster on July 26, 1985 at Nagano city. Natural Disaster Research Report No. 399

26, NRCDP, Tsukuba, Japan, 45pp (in Japanese). 400

Oyagi, N., Terajima, H., Moriwaki, H., 1977. Report on the Fukuchi-Nukeyama 401

Landslide disaster in Ichinomiya Town, Hyogo Prefecture and debris-flow 402

disaster in Syodo Island, Kagawa Prefecture, due to Typhoon No. 17, 1977. 403

Natural Disaster Research Report No. 13, National Research Center for Disaster 404

Prevention (NRCDP), Tsukuba, Japan, 68pp (in Japanese). 405

Roering, J.J., Kirchner, J.W. Dietrich, W.E., 2005. Characterizing structural and 406

16

lithologic controls on landsliding: Implications for topographic relief and 407

landscape evolution in the Oregon Coast Range, USA. Geological Society of 408

America Bulletin 117, 654–668. 409

Scheidegger, A.E., 1973. On the prediction of reach and velocity of catastrophic 410

landslides. Rock Mechanics 5, 231–236. 411

Shimizu, F., Oyagi, N., 1985. Landslide Maps, Series 3 ”Hirosaki and Fukaura”. 412

Technical Note No. 96, NRCDP, Tsukuba, Japan. 413

Shimizu, F., Oyagi, N., Inokuchi, T., 1984. Landslide Maps Series 2 ”Akita and 414

Ojika”. Technical Note No. 85, NRCDP, Tsukuba, Japan. 415

Shimizu, F., Yagi, H., Higaki, D., Inokuchi, T., Oyagi, N. 2003. Landslide Maps, 416

Series 16 “Nagano”. Technical Note No. 238, NIED, Tsukuba, Japan. 417

Shimizu, F., Oyagi, N., Miyagi, T., Inokuchi, T., 2004. Landslide Maps, Series 17 418

“Nagaoka and Takada”. Technical Note No. 244, NIED, Tsukuba, Japan. 419

Shimizu, F., Inokuchi, T., Oyagi, N., 2005a. Landslide Maps, Series 23 “Wakayama 420

and Tanabe”. Technical Note No. 271, NIED, Tsukuba, Japan. 421

Shimizu, F., Inokuchi, T., Oyagi, N., 2005b. Landslide Maps, Series 24 “Himeji”. 422

Technical Note No. 277, NIED, Tsukuba, Japan. 423

Yonetani, T., Moriwaki, H., Shimizu, F., 1983. Report on the debris-flow disasters in 424

Isshi County, Mie Prefecture and Wada Landslide disaster in Nishiyoshino 425

Village, Nara Prefecture, due to Typhoon No. 10, 1982. Natural Disaster 426

Research Report No. 22, NRCDP, Tsukuba, Japan, 70pp (in Japanese). 427

Yoshimatsu, H., Abe, S., 2006. A review of landslide hazards in Japan and assessment 428

of their susceptibility using an analytical hierarchic process (AHP) method. 429

Landslides 3, 149–158. 430

431

17

Figure legends 432

433

Fig. 1. Section of landslide map ‘Hachimantai’ (Shimizu et al., 1984). Crowns are 434

represented by thick solid lines, and displaced masses by thick broken lines. 435

436

Fig. 2. Locations of investigated areas. 437

438

Fig. 3. Topography and cross sections of recent landslides. (a) Sumikawa landslide 439

(modified after Hoshino and Asai, 1997; Oyagi and Ikeda, 1998). (b) Jizukiyama 440

landslide (Oyagi et al., 1986). (c) Wada landslide (Okunishi and Okuda, 1982; 441

Yonetani et al., 1983). (d) Fukuchi landslide (Oyagi et al., 1977; and Okunishi, 1982). 442

Hatched area shows residential area. Star in (a) shows the location of steam explosion. 443

Different outlines for toe of displaced mass of Wada and Fukuchi landslides have 444

been reported: (1) Okunishi and Okuda (1982), (2) Yonetani et al. (1983), (3) 445

Okunishi (1982), and (4) Oyagi et al. (1977). 446

447

Fig. 4. Number of available relic landslides (a) and relative height (b) as a function of 448

distance from recent landslides. 449

450

Fig. 5. Method for morphometric analysis of relic landslides. 451

(a) Definition of measured items. (b) Example of extrapolation for estimating foot 452

elevation. 453

454

Fig. 6. Effect of landslide size on H/L ratio. Thin broken lines show regression lines 455

for relic landslides. 456

457

Fig. 7. Effect of initial slope on H/L ratio. Broken lines show regression lines for 458

18

relic landslides. Thick solid lines show lower prediction limit with 95% probability 459

for relic landslides. 460

461

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

1

Tables 1

2

Table 1. Environmental conditions of investigated areas. 3

4

Investigated

area Main geology*

Precipitation**

(mm/yr)

Temperature**

(°C)

Altitude range

(m)

Landslide

area (%)

Hachimantai VR (Q,N)

with SR (N) 1880 6.9 1613 – 180 23.1

Nagano SR (N) with

VR(Q,N) 900 11.7 1917 – 330 13.0

Gojo SR (M) with

crystalline schist 1370 14.5 1177 – 82 5.3

Ichinomiya VR (M), SR (M) with granodiorite 1770 13.3 1730 – 61 3.7

*VR: volcanic rocks, SR: sedimentary rocks, (Q): Quaternary, (N): Neogene, (M): Mesozoic. 5

**annual mean for recent 22–30 years observed at Japan Meteorological Agency automated 6

meteorological stations (AMeDAS). The name of the referred station is same as the name of each 7

investigated area. 8

9

2

10

Table 2. Geometry of recent landslides. Area of rupture surface, Ar; relative height of 11

entire landslide, H; length of entire landslide, L; slope of rupture surface, tanθr. 12

13

Name of

landslide

Investigated

area

Ar

(km2)

H

(m)

L

(m) H/L tanθr

Sumikawa Hachimantai 0.139 350 2250 0.156 0.256

Jizukiyama Nagano 0.085 205 732 0.280 0.369

Wada Gojo 0.027 95 223 0.426* 0.494

Fukuchi Ichinomiya 0.054 152 600 0.253 0.358

*calculated from outline (2) 14

15

3

16

Table 3. The 95% lower prediction limits calculated from initial slope of recent 17

landslides and datasets of relic landslides for different D values. 18

19

Hachimantai Nagano Gojo Ichinomiya

H/L, recent 0.156 (Sumikawa) 0.280 (Jizukiyama) 0.426 (Wada) 0.253 (Fukuchi)

D = 5 km n.s. (N =5)* 0.269 (N = 25) 0.383 (N = 18) 0.229 (N = 7)**

D = 10 km 0.132 (N = 40) 0.261 (N =52) 0.384 (N = 43) 0.259 (N =38)

D = 15 km 0.149 (N = 73) 0.263 (N = 92) 0.399 (N = 93) 0.260 (N = 80)

* analysis not significant (p > 0.05), ** analysis less significant (p ~ 0.01) 20